The author also discusses the reliability and consistency of AI, suggesting that the most reliable AI solutions will be those that rely less on raw AI generated output and more on comprehensive tooling to drive automation. The author predicts that AI will be used as part of a declarative execution engine, and that this approach will be necessary for applications that require reliability. The author concludes by suggesting that grounding AI is a requirement for consistency and reliability.
Key takeaways:
- Declarative programming is a higher-order form of programming where the computer is instructed to perform a task but otherwise left to 'figure out' how to do it, as opposed to imperative programming where detailed directions are provided.
- AI/LLMs could solve the instruction language/format and the tool set problem faced by declarative systems, eliminating the need for developing a custom DSL and allowing the AI's 'intelligence' to fill in the gap for tooling.
- AI as part of a declarative execution engine can be more reliable and consistent, as the processing logic serves as a guardrail to ensure what's being done makes sense.
- While the leading AI models are very smart, they are still 'unhinged', and grounding them is a requirement for consistency and reliability, which can be achieved through a declarative processing approach.